HV5812 Driving Nixies

[NivagSwerdna]

I am toying with the idea of using a pair of HV5812 to drive six Nixie Tubes (10 digits + 2 dots)...

To save wires I would like to daisy chain the two HV5812 by connecting data_in of the downstream one to data_out of the upstream one.

The data sheet seems a bit ambiguous....

Presumably I clock each bit in using CLK (wired to both)

And then when I eventually have clocked in 40 bits I then pulse Strobe.

Do you think that would work?

[Skashkash]

That's how it's typically done.
I've cascaded hv5812s in the past to drive leds.

If you do this to drive nixies, just make sure you keep the voltage that the hv5812 actually see below the 80 or so max volts limit for that part.

You may also see some ghosting with the mux scheme shown in the schematic , but I've not looked too close at your circuit to say exactly.
 You may need to add some resistor dividers to bias the nixie pins to some intermediate voltages below the HV. Like 1/2 HV. Again, not sure.

[NivagSwerdna]

Thanks for the reply.

I'm really confused now! I have looked around and found plenty of HV5812 based Nixie designs.  In the multiplexed designs the current is increased but the multiplexing time means that average current works itself out... however the HV5812 seems good at sourcing current and rubbish at sinking it.  It seems to be that in the designs the HV5812 are attached to the nixie cathodes so are sinking.

I read somewhere that HV5812 can sink around 2.5mA, that's on the low side for driving a nixie?

Maybe I should just forget the technical details and just believe it is the right device for the job? 

[Ian.M]

It depends on the peak current you actually need.  As nearly all Nixies available are used or N.O.S. (apart from a small handful of boutique manufacturers hand-building tubes for lotsa-wonga), most SANE users drive them very conservatively so don't need high currents.  However if you are multiplexing you'll be wanting to push them a bit harder, maybe up to 5mA as the max. pulsed current is 6mA.

ZM1332 data (translated)

With the cost of N.O.S. Nixies being what it is, and multiplexing being significantly harder on them than DC operation, I would bet most HV5812 designs use DC drive with enough chips for all the cathodes of all the tubes,so aren't badly affected by its sinking limit.

If you need something beefier consider SN75468N 7x Bipolar Transistor Array, Darlington, NPN, 100 V, 500 mA, DIP.  (Basically a ULN2003 on steroids!)

Stick a 90V Zener to ground on its common (diode) pin (with a 1Meg pullup to the 170V rail to keep it biassed), and current limiting resistors in series with each collector output and it will be absolutely bomb-proof for Nixie driving.  Its also dirt cheap, so the need for 6 of them and 5 74HC595 shift registers to drive them to equal two HV5812 chips isn't a major issue.

[NivagSwerdna]

It's all a bit odd... So I was looking at a design that has a 3-way mux but a 170V through 4.7k resistor.... driving a Z570M which wants 2mA.

Now clearly I don't know the muxing logic so maybe it spends a good deal of its time turned off but... 170 intp 4k7 is 36mA!

I'll probably stick with the HV8512 (because Microchip are happy to sample me some) but tempting to just direct drive at a sensible 2.5mA.

[Skashkash]

Always some tradeoffs between direct drive and muxing with nixies.

 All in all, I find direct drive easier. You can add another hv5812 or upgrade to  hv5522. If you can get some samples.

You won't need the anode transistors, and you can PWM the blanking pin for some display dimming.

When you do your math, don't forget the voltage drop across the nixie itself.

[Ian.M]

The Z270M data I've found quotes the maintaining voltage as 140V, so  the design that was scaring you will only have about 30V at the cathodes.  If that design also used HV8512 drivers, well the worst case Output Low level is 3V (@1mA, 25 deg C) and typically its 1.5V.  As a very rough approximation you can treat it as if it had an internal 1.5K resistor, which means the peak cathode current is only going to be about 4.8mA, which is far less scary for multiplexed drive. 

[NivagSwerdna]

well the worst case Output Low level is 3V (@1mA, 25 deg C) and typically its 1.5V.  As a very rough approximation you can treat it as if it had an internal 1.5K resistor, which means the peak cathode current is only going to be about 4.8mA, which is far less scary for multiplexed drive. 

Ian, I think you are explaining the bit I still don't understand and that is confusing me.

Indeed it is HV5812 on the cathode side.

When you say "worst case output level".... is that the level that the HV5812 cannot pull lower?  So the HV5812 pin is around 3V when OFF.

And when the tube is fired then it has +170V at the anode and +30V at the cathode?

What is a good equivalence model for a Nixie tube?  Is it just a resistor (once it is conducting)? What sort of value?

[Sorry to be struggling with this]

[Ian.M]

A nixie is an over-grown mutiple electrode Neon, and you certainly cant model neons as simple resistors. Like any other gas discharge tube their characteristic has significant negative resistance regions.  However as long as you are passing a minimum current they are more or less resistive + a voltage offset of their quenching voltage.

Fortunately  LTSPICE has a parametric neon bulb model, but it goes crazy if you try to current bias 'neonbulb' below its holding current.  Even a voltage source ramping up  and a resistor feeding the tube can get pretty weird, but its likely to actually do this in real life.  I've tweaked the bulb parameters to approximate your Nixie and parametrised the simulation to make it easy to set initial and final voltages and total time (.param line):

[NivagSwerdna]

Interesting.

I'm going to work on the basis that my little tubes (I don't the the $$$ for nice big ones) only pass around 3mA and that is OK for the HV5812 to sink..

Given I want to drive both the numbers and the dots and I don't want too many components I have settled for a 2x mux scheme.

See attached

[Ian.M]

Looks like you are driving them a little 'hot' which the HV5812 may not like.
According to the data I linked nominal operating voltage is approx 145V.  Subtract that from your 170V supply and you get 25V left.  You'll loose a bit of that in the HV5812, but will still have over 20V drop across your anode resistors and nearly 5mA current.  6K8 resistors would be a better choice than 4K7 ones if you are shooting for a nominal 3mA current.   However the tubes can handle 6mA pulsed for multiplexing, and the HV5812 doesn't quote an upper pulldown current limit (only an overall dissipation limit) so its probabby best to suck it and see - get some measurements of the HV5812 Vo_low voltage and the drop across the anode resistors (to give you the current) in actual operation, check the brightness and whether the discharge is stable and only then decide whether you need to modify the anode resistor values.

[NivagSwerdna]

Looks like you are driving them a little 'hot' which the HV5812 may not like.

I'll at least double my values on the outputs of the anode drivers and work downwards after some in circuit measuring.  Really appreciate your feedback.  Thanks!
I have to draw up the power supply side and add a microcontroller and then I'll draw the PCB and send it to Elecrow... I will go quiet for a few weeks now. 

[Ian.M]

No problems.  Just bookmark the topic so you can carry on from where we've left it.   What are you proposimg for the 170V HT supply?   3 tubes on @ 3mA + it needs a bit of margin so 15mA-20mA @170V.

[NivagSwerdna]

I've done a quick review of the PSU designs in circulation... I think I'm going down this route...

https://threeneurons.wordpress.com/nixie-power-supply/

[Ian.M]

Looks sane.  However it needs some odd resistor values that could be a PITA.  I'd probably use 2x 220K for R3 and make R2 2K7 with a 1K preset in series initially set to its maximum, then trim for 170V out.

[NivagSwerdna]

My Nixie project is running at very low priority but I continue to think about it in the background...

I was wondering about actually trying to drive the tubes with constant current rather than +170V and a current limiting resistor...

I stumbled across... https://groups.google.com/forum/#!topic/neonixie-l/MxH_ppjVCa4

This sounds like a good idea although I have no idea how you would generate PDRV (being an accurate 12V less than VXX200)

Any ideas for constant current drivers for a Nixie?

[NivagSwerdna]

As above I'm trying to look at a constant current driver for the anodes... having dipped into Art of Electronics I see that the circuit by gregebert does have the same sort of look to it.  I knocked up a circuit in Tina which I had lying around and using a similar device (albeit it not a FQD7P20TM which is probably the part to use for real) I can get a circuit that delivers 3mA when 190V is applied; i.e. 10V less than the 20V voltage source.

Please can I have a hint on how I could generate -10V (accurately) less than the high voltage rail?

[Ian.M]

There's more ways of doing that than you can imagine.  The simplest would be a Zener from the gate to the +HT rail with a pulldown resistor keeping it biassed at an appropriate current.  You could also use a fixed resistor between the gate and the +HT rail and pass a known current through it from a ground referenced current source.

Then there's fancier stuff like using an isolated DC-DC converter to provide a low voltage supply that can be referenced to the +HT rail, so you can stick an OPAMP in the current control loop and servo out the offet cased by the somewhat imprecise threshold voltage.

However I see very little benefit in doing so.  The tube's operating voltage at its specified current is set by its geometry, gas fill, and cathode surface material and condition.  Any sizeable change indicates a problem with the tube - loss of vacuum, cathode poisoning or excessive electrode damage due to sputtering, and the correct course of action is to detect the change and shut down so the tube can be checked and, if its a potentially reversible condition like cathode poisoning, hopefully fixed.  Continuing to force the same current through a dying tube can do no good.
Also an active constant current circuit is inherently less reliable than a well chosen dropper resistor and if it fails, unlike a dropper, will tend to fail short circuit with a high probability of tube and driver damage.   (Dropper resistors tend to go high or open and simple replacement cures the fault with no risk of tube or driver damage.)

[NivagSwerdna]

I think the intention was to reduce overall power consumption (i.e. not waste it in the anode limiting resistor)?

[Ian.M]

It doesn't help much driving Nixies.  +HT has to be greater than the striking voltage  with sufficient margin for reliability, so you still have to drop much the same voltage when running.  All its doing is shifting the dissipation from a resistor to a semiconductor.

[NivagSwerdna]

For version 1, I'm using nixies I have lying around from an old frequency counter... they are ZM1332 (so really small)...

When laying out the 2 layer PCB I have trouble routing the tracks since the pins are so small and I need to route a few tracks between the pins...

How much clearance do you need at 170V?  Is there a way to calculate it.  (Also how about track width?)

Thanks in advance

[I decided to stick with the HV5812s, with a 62v zener)

[ebclr]

Why not 74141 ?    ( http://www.ebay.com/bhp/nixie-driver )

Interesting link

https://web.jfet.org/nixie-1/NixieTransistors.pdf

[NivagSwerdna]

I have nothing against 74141, got some of them too. In fact I also did consider using 1820-0092 which I have 10-off and they look pretty.
But I have some HV5812 in the parts box and they seem ideal.

I have 10 digits and a left and right decimal point to drive so it was always going to be a strange configuration.

(and this is only the Mk 1 prototype)

[NivagSwerdna]

I found... http://www.smpspowersupply.com/ipc2221pcbclearance.html

Seems to imply that for my coated tracks at 170V I would need a gap of 0.4mm

...that's a shame because the tightest gap I have on my board is 0.21mm.  Maybe I'll get smoke and fire! 

[ebclr]

You can put some Air gaps on the critical points

http://electronics.stackexchange.com/questions/173575/air-gap-on-isolation-space